The challenge of navigating in environments where GPS signals are unavailable, such as tunnels and underground parking structures, has been significantly addressed by a collaborative team from Wuhan University and Chongqing University. Their innovative solution, the DMDVDR (Data- and Model-Driven Vehicle Dead Reckoning) framework, utilizes a custom-designed deep neural network, AVNet, to estimate a vehicle's position with high accuracy using only the inertial sensors in a smartphone.
This system ingeniously combines artificial intelligence with classical control theory, processing raw data from a smartphone's Inertial Measurement Unit (IMU) to estimate vehicle orientation and velocity. These estimates are then integrated into an Invariant Extended Kalman Filter (InEKF) to mitigate sensor noise and drift, achieving an impressive horizontal translation error of just 0.4% in real-world tests. This performance notably surpasses that of existing solutions, marking a significant advancement in navigation technology.
The implications of this development extend beyond personal navigation, offering new possibilities for autonomous parking assistance and fleet management in environments where GPS signals are weak or absent. By presenting a scalable and cost-effective alternative to traditional in-vehicle navigation systems, the DMDVDR framework is poised to play a pivotal role in the future of smart mobility. For further information on this pioneering research, visit https://doi.org/10.1186/s43020-025-00168-7.
